voice compression

[a.helou]

I have a client who has recently established voice over ip. Apparently someone has told him that we can achieve voice compression using tcp/ip header compression. So he asked me and I'm trying to give him a satisfactory answer but I can't find any crucial result . Can you please give me a valid answer?

Thank you.

[rickb]

If you are using IP Telephony, you can use the "ip rtp header-compression" to compress the tcp/ip header.

Details: Does it really compress it? No, it just uses a smaller header with changes to initial header. Voice packets use RTP as a transport and as such have an IP/UDP/RTP header that is 40 bytes. Well a lot of that is redundant information and can be compressed down to 2 to 5 bytes. Condidering the voice package itself is small, you tremendously thin out the traffic. The lines to implement are fairly simple, see the below links for more information.

Rick

Theory:

http://www.cisco.com/univercd/cc/td/doc/product/software/ios121/121cgcr/qos_c/qcprt6/qcdlem.htm#xtocid4

Application:

http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/12cgcr/qos_c/qcpart6/qcrtphc.htm

[robrhodes]

Compression of the voice sample - the payload of the packet - is accomplished using a codec, generally g.729. Compression of the ip/udp/rtp header is accomplished using cRTP. There are caveats to enabling cRTP. Enabling cRTP will depend upon the type of link, link speed, router platform & IOS, router load, etc.

http://www.cisco.com/warp/public/788/pkt-voice-general/bwidth_consume.html

http://www.cisco.com/warp/public/105/compression-qos.html